Peter Žužek

> A minor point I see is that before by default an `accessor` constructor without a CGH was a host `accessor`. Now it is no longer the case, since it...

> The compiler should warn about it at least. > But this is what I do in all the example I show because it is shorter... I think that calling...

@intel/llvm-gatekeepers please merge. I've added https://github.com/intel/llvm/pull/14953/commits/81c469116b2c33ba1a8570ce6cc74c2d559d54a3 after approvals since we spotted this internally, but that is a documentation only fix.

@intel/llvm-gatekeepers there seems to be an issue with CI: https://github.com/intel/llvm/actions/runs/10513800450/job/29130422674?pr=14953 . Can we please merge this since the issue is blocking other work, the CI passed before my NFC change,...

I don't think we should be replacing `throw`s with `assert`s in `id` and `range` classes in this PR, that seems out of scope. Also friendly ping @intel/llvm-reviewers-runtime for reviews.

We're looking into this internally. We've noticed issues with this test in the past - there might be an issue with rounding on different devices.